Investigation of added utility of nonlinear techniques in rescaling soil moisture datasets

Download
2019
Hesami Afshar, Mahdi
Soil moisture plays a key role in weather forecasting, hydrologic modeling, climate change studies and water resource management. There are multiple ways to estimate this essential variable (i.e., remote sensing, modeling, station-based observations) and clear benefits associated with merging independent estimates. However, the time series of these products generally contain systematic differences that must be removed through rescaling before the application of data merging approaches (e.g., data assimilation and data fusion). In this study, the added utility of nonlinear rescaling methods relative to linear methods in th e framework of creating a homogenous soil moisture time series has been explored. The performances of 18 linear and nonlinear rescaling methods are evaluated in two different case studies of: 1) rescaling the AMSR-E LPRM soil moisture dataset to station-based watershed average soil moisture (WASM), and 2) fusing of four different soil moisture products (ASCAT, AMSR-E LPRM, API, and NOAH) via a naive data fusion scheme and multiple rescaling approaches. Accordingly, experiments are performed using various rescaling methods, where the rescaled and fused datastes are validated using observations obtained over four United States Department of Agriculture (USDA) Agricultural Research Service (ARS) watersheds, which are frequently used in the validation efforts of the soil moisture satellite missions. The results of a total of 18 different methods show that the nonlinear methods improve the correlation and error statistics of the rescaled product compared to the linear methods. In general , the method that yielded the best results using training data (ELMAN ANN) improved the validation correlations, on average, by 0.052, whereas JORDAN ANN and MARS, yielded correlation improvements of 0.038 and 0.01, respectively. On the other hand, results related to the validation of fusion of products obtained via a smooth-deviance decomposition rescaling technique, show, on average, a correlation improvement of 0.03, compared to the other widely implemented simple linear rescaling approaches. The overall results show that a large majority of the similarities between soil moisture datasets are due to linear relations; however, nonlinear relations clearly exist, and the use of nonlinear rescaling methods or implementation of linear methods with a proper rescaling approach clearly improves the accuracy of the rescaled product. Additionally, the selection of the reference dataset from higher quality datasets in the rescaling steps results in considerably increased fused product accur acy.
Citation Formats
M. Hesami Afshar, “Investigation of added utility of nonlinear techniques in rescaling soil moisture datasets,” Ph.D. - Doctoral Program, Middle East Technical University, 2019.